Gesture-based vehicle access system with virtual keypad

Abstract

Access to a vehicle is controlled by detecting gestures pointed at a virtual keypad. A setup mode preconfigures a sequence of touchpoints as a security key, whereby an administrator performs a timed series of distinct gestures according to a selected number and position of touchpoints detected by the vehicle using captured real-time images. Touchpoints during a plurality of retries provide respective sets of data points for each of the distinct gestures in the series. A respective deviation contour is determined for each respective set and then expanded to reflect valid ranges for respective touchpoints corresponding to the sequence of touchpoints.The extended deviation contours are assumed to define the sequence of touch points for the safety closure if there is no overlap between extended deviation contours to complete the setup mode.

Description

FIELD OF TECHNOLOGY

[0001] The present invention relates generally to a camera-based security system and, more particularly, to defining a virtual keypad by a user for entering a security key. GENERAL STATE OF THE ART

[0002] Motor vehicle access and security systems are designed to prevent unauthorized access to a passenger compartment and storage compartments, as well as to restrict access to certain vehicle functions, such as starting and / or engaging a powertrain to propel the vehicle. Physical keys and wireless remote key fobs have been used as a basis for controlling access based on physical possession of a key device. Some remote key devices operate such that when a button on the remote key is pressed, the device sends a code to the vehicle to instruct the vehicle to unlock a locked closure (e.g., a door, tailgate, or trunk lid).Passive Entry Passive Start (PEPS) remote keys may include a transponder that follows a challenge / response protocol to unlock a door when a user grasps the door handle, operates or presses a button, or approaches the vehicle within a predetermined distance.

[0003] Another type of entry system known as Phone-as-a-Key (PaaK) has been introduced, where users use their smartphone to unlock a vehicle. These systems can work in a similar way to a remote key, but can typically be connected using Bluetooth. ® Low Energy (BLE), Ultra-Wide Band (UWB), NFC or other wireless technologies for mobile devices to communicate with the vehicle.

[0004] Other types of security systems are known that do not require an authorized user to possess a key, remote key fob, or telephone to access a vehicle. For example, keypad-based systems are known in which a keypad positioned on the exterior of a vehicle can be used to unlock the vehicle based on a secret numeric code entered on the keypad, or to lock the vehicle based on a secret numeric code or a publicly known code. Placing a keypad on an exterior surface of a vehicle incurs associated costs for hardware, wiring, installation, and warranty.

[0005] To maintain the functionality of a keypad while avoiding some of the costs, a virtual keypad can be created where user keystrokes are detected using an alternative sensor system, such as an image sensor. For example, a temporary keypad was projected onto a vehicle window or other surface to define the positions where the user should tap the virtual keypad to enter specific digits into a security key, and the touch of individual keys was optically detected.

[0006] To provide secure access, a camera for detecting a user's biometric data, such as facial recognition, has also been proposed. Another alternative involves optically detecting hand gestures, such as pointing a number of fingers or using other gestures (e.g., sign language), to convey a series of distinct "digits" within a security code. Because biometric data must first be captured in advance for each specific authorized user, without a keypad, a vehicle user / owner would lose the ability to share a secret numeric code with a person they wish to provide access to the vehicle if the user / owner is unavailable or if their biometric identification is unsuccessful.While hand gestures may be shareable, setting up and executing predefined gestures may require an impractical level of user training or may be subject to low detection accuracy due to a wide range of confounding variables in gesture execution, detection, and classification. SUMMARY OF THE INVENTION

[0007] The present invention enables an administrator (e.g., owner / user of a vehicle) to customize a virtual keypad constructed from arbitrary touchpoints on one or more surfaces on or near a vehicle, while ensuring that a sequence from which a security code is constructed is sufficiently diverse and repeatable to provide reliable recognition in use. Features employed as "anchor points" in a sequence may include pre-existing features on or near the vehicle (e.g., a badge, a door handle, a trim piece, a junction of two parts, a bend in the metal skin, or points on the ground) or a decal, sticker, or other material applied to the vehicle by the administrator.This eliminates the need to force the administrator to learn and then select from predefined gestures. Instead, a desired sequence can be performed and monitored in a way that ensures the end result is acceptable in practice.

[0008] In one aspect of the invention, a vehicle access system includes an image sensor configured to capture real-time images of the vehicle according to a predetermined field of view. A locking device is configured to selectively provide access to the vehicle. A controller is configured to track a plurality of touch gestures of a user external to the vehicle to identify a sequence of touchpoints that encodes a security key and operate the locking device to provide access when the user validly performs the sequence of touchpoints that encodes the security key.The controller is configured to preconfigure the sequence of touch points in a setup mode in which the controller detects an administrator's first performance of a timed series of different gestures according to a selected number and location of touch points on a surface on or near the vehicle in the predetermined field of view. The controller prompts the administrator for a plurality of retries of the series of different gestures and then detects respective touch points during the retries to collect respective sets of data points for each of the different gestures in the series. The controller determines a respective deviation contour for each respective set and then expands the respective deviation contours to reflect valid regions for respective touch points.The controller assumes the extended deviation contours as defining the safety key if there is no overlap between extended deviation contours. BRIEF DESCRIPTION OF THE DRAWINGS Fig. 1 is a schematic block diagram illustrating a vehicle and a safety system of the invention. Fig. Figure 2 is a perspective view of a vehicle with image sensors and suitable surface features for guiding a user to the touchpoints of a sequence. Fig. 3 is a partial perspective view of another vehicle with an image sensor and additional surface features for guiding a user to the touchpoints of a sequence. Fig. Figure 4 is a plan view showing a number sticker designed to be applied to a vehicle surface to provide touchpoint features. Fig. 5 is a plan view of a graphic decal designed to be applied to a vehicle surface to provide touchpoint features. Fig. Figure 6 is a data diagram illustrating detected touch points during retry attempts to develop a security code. Fig. 7 represents a set of data points and an arithmetic mean. Fig. 8 represents a standard deviation contour in the x- and y-direction for the data set Fig. 7. Fig. 9 represents a standard deviation contour drawn as an ellipse for the data set from Fig. 7 is determined. Fig. 10 shows a variety of extended deviation contours where unwanted overlap occurs. Fig. 11 is a flowchart illustrating a preferred method for configuring a security code using a virtual keypad. Fig. Figure 12 is a flowchart illustrating a process for determining deviation contours and comparing the expanded deviation contours. Fig. 13 is a front view illustrating a display panel in a vehicle on which a predetermined sequence of touch points is displayed that encodes a security key. DETAILED DESCRIPTION OF PREFERRED EMBODIMENTS

[0009] The present invention may utilize a camera sensor to collect input data representative of touch points. The invention may be implemented in any vehicle having interior and / or exterior cameras connected to a vehicle controller with image processing capability to detect gestures that involve pointing a finger or other body part at user-selected touch points. These user-defined gestures may involve touching various points on a decal / sticker or visually identifiable locations on existing landmarks, which may be parts of the vehicle or its surroundings. This decal could be a sports logo, a bumper sticker, or the like.Anything a user can reliably touch in the same location can be the basis for a virtual keypad, and security can be improved because the selected features a user targets to generate a security code may bear little or no resemblance to a traditional keypad, making it impossible to guess the security code. Features a person could reliably touch might include the top of a tailgate, a bumper, a window frame, or appliqués on a vehicle, such as a badge.

[0010] Some embodiments may use an external sound transducer (e.g., speaker and / or microphone) to allow a user to initiate the learning process through spoken commands. Vehicle-generated messages may recommend locations for touch points if the customer is unsure of which locations to use. The vehicle messages offering suggestions for good locations may be presented on a touchscreen display panel, spoken aloud, or sent wirelessly to a user's mobile device. A vehicle controller may evaluate whether a sequence of gestures / touch points is sufficiently repeatable and distinguishable to be used as a security code, which depends on a user's ability to touch the same points with sufficient accuracy as seen within the field of view of the image sensor(s).In particular, the closer the touch points are to each other, the greater the precision of the user's touch actions must be to achieve reliability. Representative images (e.g., images) of each gesture can be stored in the vehicle's control system for display to a user to remind them of a security code. The stored sequence can be accessed later, for example, from within the vehicle on the touchscreen display or via an app on a mobile device.

[0011] When a decal / sticker (i.e., any graphic or textual film that may be applied to a vehicle surface) is used, it may be preferable for image detection that the decal be transparent for aesthetic reasons and / or to allow an interior camera to detect the touch points. In particular, training (i.e., programming) for a desired sequence may require that the administrator not wear gloves or that the training be based on gestures using a pencil, stylus, pointer, or the corner of a hand-held card. Images acquired during training (as well as during normal use) may be decoded to track the gestures using pattern recognition, machine learning, or artificial intelligence systems as are known in the art.

[0012] To initiate training, an administrator (i.e., an authorized user, such as a holder or owner of a physical or software authentication key) can start the learning process by speaking a command into an external microphone. In response, a vehicle controller requests the administrator to initiate a touch / gesture within the camera's field of view and hold the touch point for a predetermined time (e.g., 3 seconds, but less than 10 seconds). Once the gesture is recorded, the vehicle controller confirms it and requests the next gesture. A plurality of such gestures can be detected and recorded. For a final gesture, the administrator can hold the gesture for a longer period of time (e.g., 10 seconds). The vehicle controller announces that the sequence is being recorded and then requests that the sequence be repeated for a plurality of retries (e.g., 10 times).The purpose of the retries is to determine whether the gestures can be reliably recognized (e.g., whether there is sufficient separation between touch points). In some embodiments, the retries are evaluated to determine whether there is an overlap between gestures targeting different touch points in the sequence. An overlap can be defined as conflicting gestures that are within a certain number of standard deviations of the gestures that targeted corresponding touch points. If an overlap exists, the vehicle controller can make suggestions on how to improve the distinguishability of each gesture (e.g., by displaying images of the overlaps and indicating where the distinguishability of each gesture can be improved).

[0013] In some embodiments, other types of user actions (e.g., sounds, spoken words, or movements) may also be incorporated into a sequence of gestures to further enhance the security of the requests. For example, a word or phrase may be configured to occur at a selected position within the sequence, which can be detected and recognized by the vehicle controller using speech recognition to significantly improve security.

[0014] Related to Fig. 1, a vehicle 10 includes on-board systems 11 that support a vehicle access function that includes authenticating a user 12. During training, the user 12 is an administrator who can be recognized based on another security key, such as a wireless key fob or a phone-as-a-key (PaaK). During normal use (i.e., after training has taken place), the user 12 can be any person who knows the trained sequence of actions (e.g., touch points, gestures, or spoken words) that includes a security key that allows access to the vehicle 10 (e.g., unlocking a door, turning on electronic systems, or starting a powertrain).

[0015] The vehicle 10 includes one or more image sensors, such as cameras 14-16, employed to obtain a respective field of view (FOV) directed toward the exterior of the vehicle 10, such as the FOV 17 of camera 16. The image sensors may include, for example, CMOS visible light sensors or LiDAR sensors. Some on-board systems 11 are interconnected by a communication bus 18, allowing an authentication controller 20 to receive data from the cameras 14-16 and to exchange data and commands with a driver interface module 21 and other accessory modules 22. The authentication controller 20 is also coupled to a door locking / unlocking mechanism 23, activation inputs 24 (e.g., door handle activation sensors), interface devices 25 (e.g., speakers, microphones, car horn, sounders, and / or exterior lights), and a powertrain controller 26.The authentication controller 20 includes a database 24 in which security codes and associated images are stored after training. The user 12 may carry a mobile device (e.g., a smartphone) 25 that communicates with the interface 21 to support the operation of the training mode or to provide a display screen for a user during subsequent use.

[0016] Upon approaching the vehicle 10, the user 12 can act as the administrator, who can initiate the training of a security code, or can act as a regular user to execute an existing security code. Both actions could be initiated, for example, by tapping or activating a door handle or by speaking a command word or phrase. The smartphone 25 can be used in particular in connection with activating and executing the training mode.

[0017] When using a decal / sticker to create a virtual keypad, the decal / sticker can first be placed on the vehicle within a camera's field of view. Fig. 2 illustrates a vehicle 30 with a decal 31 affixed to a window 32 within view of an interior camera (not shown). The decal 31 may be partially transparent and is depicted with a drawing of a sailboat to provide multiple landmarks for use in defining a security code. The vehicle 30 also shows an exterior camera 33 (e.g., on a tailgate of a pickup truck) with a field of view that allows the camera 33 to capture real-time images of a brand badge 35 and design surface features (e.g., indentations) 36, all of which may provide landmarks to be incorporated into the gestures that define a security code.

[0018] Fig. 3 illustrates a vehicle 37 with an exterior camera 38 mounted in a sideview mirror 39. A field of view of the camera 38 includes a door panel having a decal 40 with landmark features, such as button graphics, applied thereto. The camera 38 may be integrated with a light source that projects a landmark image 41 onto a ground surface proximate the vehicle 37. The image 41 includes symbols 42A, 42B, and 42C to guide a user in performing gestures for a security code. The pointing gestures could be performed, for example, by pointing with a user's foot.

[0019] Fig. 4 illustrates a decal 43 that may include a backing surface carrying an adhesive for attachment to a vehicle surface. A front surface of the decal 43 displays a plurality of visible features that the user can associate with a sequence of touch points, the features including a virtual button area 44 and a virtual button area 45 that can be tapped with a user's finger 46. To make a sequence more memorable, button areas may be marked using numbers, letters, words, images, symbols, logos, colors, or other recognizable markings. Fig. 5 illustrates the sailboat decal 31 in more detail. The features that a user can select as a touch point include a flag 47, a bow 48, and a stern 49. Anything that a user can remember and that can guide a touch gesture with sufficient distinctiveness and repeatability could be selected.

[0020] To evaluate such discriminability and repeatability, the user must perform a large number of repeat trials for any given sequence. Fig. Figure 6 depicts a data space 50 having coordinates defined according to the field of view of the camera(s) capturing real-time images of the corresponding gestures. For each separate touch point in a series of different gestures, a corresponding set of data points is collected. A set of data points 51 (corresponding to a particular gesture in a selected series of different gestures, all directed toward a respective position selected by the administrator) is represented by small x's. A set of data points 52 is represented by small o's, and a set of data points 53 is represented by small boxes, each set corresponding to a respective gesture in the series. Each data point in a respective set is an instance of a replication of a respective touch point targeted by the administrator / user.The distribution of data points within a set for a respective touch point and the separation between the respective sets can be analyzed to determine whether the gestures can be reliably detected and classified.

[0021] In particular, a deviation contour is determined for each respective set of data points as a region that may contain valid instances of the touch points for a respective gesture. In some embodiments, the deviation contour is characterized according to a deviation with respect to an expected value calculated from the data points. Preferably, the deviation may comprise a standard deviation, and the expected value may be an arithmetic mean (e.g., average). Alternatively, the deviation may comprise a contoured boundary circumscribing all of the data points. For example, the contoured boundary may be defined as a circle with a center located at an average position of the data points and a radius equal to a maximum of the distances of the individual data points from the average.

[0022] Fig. Figure 7 depicts the data points of set 51 plotted in two dimensions, with an expected value 54 determined as the arithmetic average of the x-coordinates and the y-coordinates. The standard deviation is useful as a measure of the degree of dispersion of a data set. Fig. Figure 8 represents a deviation contour 57 according to a calculation of the standard deviation (σ) in X and the standard deviation in Y with respect to the (mean) expected value 54. The standard deviations in X and Y are the square roots of their variances, as follows: σX=∑(X−μ)2N σY=∑(Y−μ)2N where µ is the mean, N is the population size of the data set, X is the set of X coordinates for the population of data points, and Y is the set of Y coordinates for the population of data points. The contour 57 is a rectangle centered on the expected value 54, with a side 55 of ±σ x (ie a width of 2σ x) and a page 56 of ±σ y (ie a height of 2σ y ). Alternatively, the standard deviation contour can be calculated as a standard deviation ellipse (SDE) based on the X and Y standard deviations as follows: SDE=σX2+σY2.

[0023] The SDE is centered on the mean to provide a deviation contour 58, which in Fig. 9 is shown.

[0024] The deviation contour reflects a range of scatter across which a user's attempts to direct a gesture toward a corresponding feature have fallen. To provide sufficient separation between different touch points within a series defining a security code to ensure that respective gestures are distinct, the deviation contours are extended to provide a buffer margin between them. The extended contours represent valid regions for detecting each respective touch point in the security code. To provide a valid series of gestures, the extended contours must not overlap, which would introduce ambiguity when determining which gesture is intended for a particular touch point in a security code.

[0025] Fig. Figure 10 illustrates a plurality of deviation contours and respective extended contours obtained after a plurality of retries during a setup mode for creating a security key. Standard deviation contours 54, 61, and 63 are shown, with each rectangular contour having a size of ±1σ x and ±1σ y , of the X-coordinates and the Y-coordinates of each respective set of data points. In this example, each deviation contour is expanded by multiplying by a predetermined factor of 2, so that each expanded deviation contour has a size of ±2σ x and ±2σ yas represented by the extended contours 60, 62, and 64. The extended contours 62 and 64 enclose an overlap 65 as a result of the extension. This indicates that the selected positions for respective touch points in the sequence are not sufficiently separated and distinct. Therefore, the performed series of gestures is unacceptable. Thus, an administrator would be prompted to modify the desired series of touch points. Preferably, an assignment of the overlap would be communicated to the administrator so that the problem can be avoided during a subsequent attempt in setup mode.

[0026] Fig. 11 illustrates a preferred method of the invention, wherein an administrator initiates setup mode at step 70. Entry into setup mode may be triggered by the administrator, for example, by issuing a spoken command outside the vehicle or by entering a command via a remote control app on the administrator's mobile device or on a touchscreen display in the vehicle. At step 71, the vehicle controller acknowledges entry into setup mode and may then provide guidance to the administrator via audio prompts and / or graphic and text displays on the vehicle display panel or mobile device.When ready, the administrator performs a first execution of a time series of different gestures according to a selected number and location of touch points on a surface on or near the vehicle that is within a predetermined field of view of the vehicle's image sensors. At step 72, the vehicle controller signals that it is ready to receive a gesture. At step 73, the administrator maintains a gesture on a particular touch point for a predetermined time (e.g., 3 seconds). For example, the administrator may maintain an index finger on a particular character or digit on a designed decal applied to a vehicle exterior surface. At step 74, the vehicle controller checks to determine whether a gesture has continued for the predetermined time.If not, an error message may be generated at step 75 to inform the administrator, and a return is made to step 72 to prompt the administrator to perform a next gesture. If a gesture is detected at step 74, the position indicated by the gesture is recorded at step 76. At step 77, the vehicle controller checks for a completion signal. A completion signal may involve the administrator continuing a particular gesture for an even longer period of time (e.g., 10 seconds). Alternatively, a completion signal could be generated by a spoken command or by manually entering a completion command on the administrator's mobile device or on the vehicle's touchscreen display panel. If no completion signal is present, a return is made to step 72 to obtain a next gesture in the series of gestures.While performing the series of gestures, the vehicle control system may offer the administrator various suggestions or other guidance as needed.

[0027] Once a completion signal is detected, a plurality of retries are requested by the vehicle controller at step 78. Sufficient retries are repeated so that a sufficient population data size is obtained to allow confirmation of the distinctiveness and repeatability of the sequence of gestures (e.g., 10 retries). In response, the administrator performs the sequence the predetermined number of times while the vehicle controller collects data points for the consecutive retries. At step 80, the vehicle controller calculates deviation contours (e.g., standard deviation contours) and then expands the contours as described above. At step 81, a check is performed to determine whether any of the expanded deviation contours overlap each other.If an overlap is detected at step 81, any areas of overlap may be displayed to the administrator at step 84, and suggested revisions to mitigate the overlap may be provided via the user interfaces (e.g., text-to-speech, vehicle display panel, or mobile device). If no detected overlap exists between extended deviation contours, the extended deviation contours are assumed to define the sequence of touchpoints for the security key at step 82 by storing the sequence. At step 83, the vehicle controller may generate a map of the stored sequence and store it for use as feedback and / or for display in response to subsequent user requests to recall the security key. This completes setup mode.Later, when the vehicle is locked or functions are otherwise restricted, user access attempts into the vehicle can be tracked. A user can initiate an unlocking process (e.g., by activating a door handle) and then perform the variety of touch gestures outside the vehicle while being monitored by the vehicle's cameras. If the user performs the sequence of touch points that correctly encodes the security key, a locking device inside the vehicle is triggered to provide access to the vehicle.

[0028] Fig. 12 is a flowchart further illustrating a preferred method for determining standard deviation contours and extended contours. At step 88, X and Y coordinates for the tangency points within each respective data set are compiled. At step 89, expected values ​​are determined for each respective data set. For example, the arithmetic means for the X and Y coordinates are calculated, with the respective mean X and Y values ​​defining each expected value. Based on the expected value of each data set, at step 90, a range of validity is determined based on the standard deviation, maximum values, or other characterization of the dispersion around the expected value for each data set. At step 91, the deviation contours are extended to larger ranges using a buffer margin.Preferably, a standard deviation contour may be expanded by multiplying the respective standard deviations by a predetermined factor in a range of 1.5 to 3.5. More preferably, the predetermined factor may be between 2 and 3. In an embodiment where a deviation contour is based on a maximum distance from the mean (rather than the standard deviation), the expansion may also include multiplication by a predetermined factor selected to provide a desired buffer margin. At step 92, a check is performed to determine if expanded regions overlap. If they overlap, the proposed sequence of touch points is rejected at step 93. Otherwise, it is accepted at step 94.

[0029] In Fig. Figure 13 depicts an example of a touchscreen display panel 95 in a vehicle. The display panel 95 displays images to assist in the generation and / or retrieval of gestures in a security key. Images on the display panel 95 may be based on images captured by the vehicle's interior and / or exterior cameras. Thus, they may include, for example, a captured image of a decal applied to a vehicle's exterior surface. Fig. Figure 13 shows a decal 96 of a sailboat in which a plurality of touch points have been added according to a sequence previously saved in setup mode. The markings 97 indicate the sequence order for the gestures that form the security key.

[0030] According to one embodiment, in the setup mode, the controller expands the respective standard deviation contours by multiplying respective standard deviations by a predetermined factor in a range of 1.5 to 3.5.

[0031] According to one embodiment, the step of expanding the respective standard deviation contours comprises multiplying the respective standard deviations by a predetermined factor in a range of 1.5 to 3.5.

[0032] According to one embodiment, the first execution comprises one or more non-last held gestures each lasting a first predetermined detection time at a respective touch point, and a last held gesture lasting a second predetermined detection time at a respective touch point, wherein the second predetermined time is longer than the first predetermined time.

[0033] According to one embodiment, the vehicle includes a display panel visible in an interior of the vehicle, the method further comprising the step of: turning on the display panel in response to a user request for the setup mode to show images depicting the area and the sequence of touch points.

[0034] According to one embodiment, the vehicle includes a display panel visible in an interior of the vehicle, the method further comprising the steps of: rejecting the extended deviation contours defining the security key when there is an overlap between extended deviation contours; turning on the display panel to show images depicting the overlapping extended deviation contours; prompting the administrator to perform a revised series of different gestures to avoid the overlap.

Claims

[1] Vehicle access system for a vehicle, comprising: an image sensor configured to capture real-time images of the vehicle according to a predetermined field of view; a locking device configured to selectively provide access to the vehicle; and a controller configured to track a plurality of touch gestures of a user outside the vehicle to identify a sequence of touch points that encodes a security key and to operate the locking device to provide access when the user validly performs the sequence of touch points that encodes the security key, the controller configured to preconfigure the sequence of touch points in a setup mode in which the controller: (A) detecting a first execution by an administrator of a timed series of different gestures according to a selected number and position of touch points on a surface on or near the vehicle in the predetermined field of view; (B) requires the administrator to attempt a series of different gestures a number of times; (C) detecting respective touch points during the repetition trials to collect respective sets of data points for each of the different gestures in the series; (D) determines a respective deviation contour for each respective set; (E) the respective deviation contours are extended to reflect the validity ranges for the respective contact points; and (F) assumes the extended deviation contours as defining the security key if there is no overlap between extended deviation contours. [2] The vehicle access system of claim 1, wherein in the setup mode, the controller determines each respective deviation contour for each respective set according to a respective expected value for each respective set. [3] A vehicle access system according to claim 2, wherein each respective expected value comprises a respective arithmetic mean for each respective set. [4] A vehicle access system according to claim 3, wherein each respective deviation contour comprises a standard deviation contour around the respective arithmetic mean. [5] The vehicle access system of claim 4, wherein in the setup mode, the controller expands the respective standard deviation contours by multiplying respective standard deviations by a predetermined factor in a range of 2 to 3. [6] A vehicle access system according to claim 3, wherein each respective deviation contour comprises a respective region around the respective arithmetic mean having a diameter corresponding to a maximum deviation in each respective set of data points from the respective arithmetic mean. [7] The vehicle access system of claim 1, further comprising a decal applied to the surface of the vehicle within the predetermined field of view, the decal including visible features that the user associates with the sequence of touch points. [8] The vehicle access system of claim 1, wherein during the first execution, the controller (1) detects one or more non-last held gestures each lasting a first predetermined detection time at a respective touch point, and (2) detects a last held gesture lasting a second predetermined detection time at a respective touch point, the second predetermined time being longer than the first predetermined time. [9] The vehicle access system of claim 1, further comprising a display panel visible in an interior of the vehicle, wherein the controller is configured to turn on the display panel in response to a user request for the setup mode to show images depicting the area and the sequence of touch points. [10] The vehicle access system of claim 1, further comprising a display panel visible in an interior of the vehicle, wherein the controller is further configured, in the setup mode, to (i) reject the extended deviation contours as defining the security key when there is an overlap between extended deviation contours; (ii) turn on the display panel to show images depicting the overlapping extended deviation contours; and (iii) prompt the administrator to perform a revised series of different gestures to avoid the overlap. [11] A method for controlling access to a vehicle, the vehicle including an image sensor configured to capture real-time images according to a predetermined field of view of the vehicle and a locking device configured to selectively provide access to the vehicle, the method comprising the following steps: Entering a setup mode to preconfigure a sequence of touchpoints as a security key; Performing a first execution of a timed series of different gestures by a registrant according to a selected number and position of touch points on a surface on or near the vehicle in the predetermined field of view; Identifying the positions of the touch points using the captured real-time images; Performing a plurality of repetition attempts of the series of different gestures by the registrant; Identifying the positions of the touch points during the repetition trials using the captured real-time images to collect respective sets of data points for each of the different gestures in the series; Determining a respective deviation contour for each respective set; Expanding the respective deviation contours to reflect validity areas for respective contact points corresponding to the sequence of contact points; Accepting the extended deviation contours as defining the sequence of touch points for the safety closure if there is no overlap between extended deviation contours to complete the setup mode; When access to the vehicle is restricted, tracking a plurality of touch gestures from a user outside the vehicle to identify the sequence of touch points that encodes the security key, and operating the locking device to provide access to the vehicle when the user validly performs the sequence of touch points that encodes the security key. [12] The method of claim 11, wherein the step of determining each respective deviation contour for each respective set comprises determining a respective expected value for each respective set. [13] The method of claim 12, wherein each respective expected value comprises a respective arithmetic mean for each respective set. [14] The method of claim 13, wherein the step of determining each respective deviation contour comprises determining a standard deviation contour about the respective arithmetic mean. [15] The method of claim 14, wherein the step of expanding the respective standard deviation contours comprises multiplying the respective standard deviations by a predetermined factor in a range of 2 to 3.

Images